Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Detailed replication explanation (e.g. locks)

Re: Detailed replication explanation (e.g. locks)

From: Merle Martell <martell_merle_at_cargill.com>
Date: 1997/09/11
Message-ID: <34185CD0.743F@cargill.com>#1/1

Brian Rasmusson wrote:

>snipped

> I would like to know when locks are set and freed, how many rows are
> locked when and why, and when the commit is issued.
> snipped
> So.. If a large batch job updates e.g. 100.000 records on the master
> site, how are they actually replicated to the site? Does Oracle make one
> huge transaction (i have seen a refresh job run out of rollback space on
> other occasions) and a single commit, or does it commit rows one by one?
> Does it lock the entire table, all rows to be updated or one row at a
> time?

In Oracle 7.3 and older, Oracle takes a table level lock. Reads can take place, but not inserts/updates/deletes. Your snapshot users need to wait until the 100.000 rows have been applied.

I don't remember if this changes in Oracle8 or not. Does anyone else know?

mmartell_at_cargill.com Received on Thu Sep 11 1997 - 00:00:00 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US